train_results.json

{
    "epoch": 5.971724787935909,
    "num_input_tokens_seen": 9124912,
    "total_flos": 3.904530641814815e+17,
    "train_loss": 1.665087498799719,
    "train_runtime": 5809.5562,
    "train_samples_per_second": 4.383,
    "train_steps_per_second": 0.136
}